New SEC Chair, Bitcoin, xAI Supercomputer, UnitedHealth CEO murder, with Gavin Baker & Joe Lonsdale
All-In PodcastGavin Baker, Joe Lonsdale, Paul Atkins, Gary Gensler, Michael Saylor, Jason Calacanis, David Friedberg
Major Personnel and Administrative Changes
- David Sacks has been appointed as the White House "AI and Crypto Czar," tasked with guiding U.S. administration policy in artificial intelligence and cryptocurrency.
- Sacks will also lead the Presidential Council of Advisors on Science and Technology.
- Paul Atkins replaces Gary Gensler as the new SEC Chair; Atkins is characterized as a former SEC commissioner under Bush 2 with a pro-crypto, fair-regulation approach.
- David Friedberg notes the irony in Trump simultaneously advocating for 100% tariffs on BRICS nations using alternative currencies while simultaneously promoting Bitcoin adoption.
- Chris Wright is nominated as Energy Secretary and is identified as pro-nuclear and pro-cheap energy.
Economic Policy and Deregulation Consensus
- Hosts and guests express a consensus that U.S. economic growth is currently constrained by excessive bureaucracy, regulations, and inefficiency.
- The administration's strategy is described as removing regulatory barriers to unlock GDP growth, aiming for 4-5% growth to manage the debt-to-GDP ratio.
- Gavin Baker compares the U.S. situation to Satya Nadella's tenure at Microsoft, suggesting that simply "stopping dumb things" (i.e., deregulation) will yield significant market gains.
- Joe Lonsdale describes the current federal bureaucracy as comparable to a bankrupt company kept alive by constant government funding, citing a lack of accountability and hard hiring tests.
- A key proposed reform is a data-driven regulatory system where rules must defend their existence, potentially including automatic sunset clauses unless renewed through a difficult, evidence-based process.
Energy Sector and Strategic Competition with China
- Participants identify the massive gap in electricity production capacity between the U.S. and China as a critical national security and economic threat.
- China is adding 6 terawatts of capacity (from 2 to 8), while the U.S. is currently adding only 1 terawatt (from 1 to 2).
- Gavin Baker and Joe Lonsdale argue that nuclear energy is the only viable solution to close this gap, citing China's aggressive rollout of Generation 4 reactors.
- Current U.S. regulations are identified as the primary blocker to nuclear and solar deployment, with Texas cited as an example where deregulation leads to rapid infrastructure build-out.
- Solar is acknowledged as having impressive cost-reduction trajectories, but its land-use requirements are described as a long-term limiting factor compared to the density of nuclear power.
Cryptocurrency and Financial Regulation
- The appointment of Paul Atkins is viewed as a shift from the "adversarial" enforcement style of the Gensler era toward a framework that provides clear rules for innovators.
- Joe Lonsdale recounts the government's "mafia-like" tactics against Facebook's Libra project, where politicians threatened to scrutinize participants to kill the initiative rather than engaging in open debate.
- David Friedberg views Bitcoin as a necessary "check" on the U.S. dollar and government excesses, arguing that competition from a pro-liberty asset is preferable to competition from authoritarian state currencies.
- A significant portion of the discussion highlights the friction between protecting retail investors and allowing wealth mobility; Joe Lonsdale and Gavin Baker criticize the current "accredited investor" thresholds as barriers that prevent average Americans from participating in private equity upside.
- Participants suggest that "sophistication tests" (e.g., courses and exams) could replace high-net-worth requirements to open private markets to a broader demographic while maintaining safety.
MicroStrategy, Bitcoin, and AI Infrastructure
- MicroStrategy's strategy of issuing convertible debt to buy Bitcoin is flagged as a high-risk trade that could become unsustainable if the debt-to-asset ratio grows too large relative to the company's operating revenue.
- Hedge funds are reportedly shorting MicroStrategy due to concerns over the leverage and the lack of collateral confidence in fixed-income markets.
- Elon Musk/XAI has successfully built the world's largest AI supercomputer, "Colossus," comprising over 100,000 NVIDIA H100 GPUs.
- This infrastructure defies previous industry consensus that connecting more than 30,000 H100s in a coherent cluster was technically impossible.
- The deployment of Colossus provides the first real-world test of scaling laws for training; if successful, it could make Grok 3 superior to other models (including OpenAI's) by January or February.
- New scaling dimensions are identified, including "test-time compute" (allowing models more time to think) and expanded context windows, suggesting the AI sector is far from hitting a performance wall.
The Assassination of Brian Thompson and Corporate Ethics
- The shooting of UnitedHealthcare CEO Brian Thompson by an individual leaving "deny," "defend," and "depose" on bullet casings is discussed as a potential protest against insurance claim denials.
- Guests condemn the violence and the online celebration of the murder, attributing the sentiment to a toxic "oppressor vs. oppressed" worldview prevalent in modern discourse.
- David Friedberg notes that while UnitedHealthcare denies claims at high rates, the industry's business model requires balancing premiums against claims; title insurance and travel insurance are cited as having worse payout ratios.
- The event raises a complex ethical question regarding CEO liability: whether executives should be personally accountable for corporate policies regarding claim denials, even if not personally negligent.
AI in Development and Productivity
- Gavin Baker reports that AI tools (specifically Cursor) enabled non-developers to build and deploy functional software applications in hackathons, suggesting a shift where human language becomes the primary programming interface.
- Current AI coding capabilities are estimated to be 70-80% effective, with the remaining work involving debugging and QA; this is projected to reach near-100% automation within 12 months.
- The "All In" podcast team notes that productivity gains from AI are already visible in startups adopting fewer employees compared to similar-sized firms from three years ago.
- Zoom and Notion are cited as examples of existing products with successful AI integration for summaries and productivity.